Learning something once is not the same as being able to recall it when it matters. I'm building a personal knowledge system around Anki to close that gap, combining careful card design, AI-assisted preparation, and a consistent review habit.
From curiosity to durable knowledge
The system connects my interest in neuroscience with practical software engineering. I use it to support Mandarin study, revisit computer science and system-design concepts, and keep a broader collection of ideas fresh. It is a learning workflow, not a claim to have invented a new memory algorithm.
The science behind the approach
The Ebbinghaus forgetting curve describes the tendency for recall to decline over time without reinforcement. It is a useful starting point, not a universal countdown for every fact or learner. A replication of Ebbinghaus's forgetting-curve research explores this pattern and its limitations.
Retrieval practice means trying to produce an answer before revealing it. In a vocabulary-learning experiment, repeated retrieval improved delayed recall more than repeated study after initial learning. That result motivates the question-first approach here, rather than simply rereading notes. See Karpicke and Roediger, 2008.
Spaced review distributes practice over time. Anki provides the scheduling layer; my system focuses on the quality, provenance, and maintainability of the material being reviewed. Anki's FSRS documentation explains how that scheduler models memory and balances desired retention with workload. I do not treat a scheduler setting as proof of my actual retention.
Mandarin and my HSK journey
Mandarin gives this work a very personal purpose. Vocabulary, pronunciation, and meaning need to remain accessible beyond a single study session. Reviewed cards give me a way to return to those details while continuing the broader work of listening, reading, and using the language. Anki supports that practice; it does not replace conversation or contextual learning.
The impact for me is continuity: material from earlier study stays in the rotation instead of disappearing when I move on to the next topic. I'll keep sharing how this develops alongside my HSK journey, without claiming an exam level or a measured learning gain I haven't documented.
How the system works
- Keep source material connected to the concepts it supports.
- Use Claude Code to help draft structured, focused questions.
- Validate the data, then review the facts and learning experience separately.
- Approve cards before exporting them to Anki.
- Revisit confusing prompts and improve the material as I learn.
For newer technical cards, a short recall target is separated from a fuller explanation. That keeps the question answerable without hiding the context. The workflow also preserves stable identifiers and revision history so improving a card does not mean losing track of where it came from.
This is a public-facing summary of a private, evolving system. No personal deck exports, source collections, or private repository are shared here.